Development of the lymphoid system is dependent on the activity of zinc finger transcription factors encoded by the Ikaros gene. Differences between the phenotypes resulting from a dominant‐negative and a null mutation in this gene suggest that Ikaros proteins act in concert with another factor with which they form heterodimers. Here we report the cloning of Aiolos, a gene which encodes an Ikaros homologue that heterodimerizes with Ikaros proteins. In contrast to Ikaros_which is expressed from the pluripotent stem cell to the mature lymphocyte_Aiolos is first detected in more committed progenitors with a lymphoid potential and is strongly up‐regulated as these differentiate into pre‐T and pre‐B cell precursors. The expression patterns of Aiolos and Ikaros, the relative transcriptional activity of their homo‐ and heteromeric complexes, and the dominant interfering effect of mutant Ikaros isoforms on Aiolos activity all strongly suggest that Aiolos acts in concert with Ikaros during lymphocyte development. We therefore propose that increasing levels of Ikaros and Aiolos homo‐ and heteromeric complexes in differentiating lymphocytes are essential for normal progression to a mature and immunocompetent state.
The Ikaros gene encodes, by alternate splicing, a family of zinc finger transcription factors which are essential for the development of the lymphoid system (Georgopoulos et al., 1992, 1994; Hahm et al., 1994; Molnar and Georgopoulos, 1994). Ikaros is first detected in pluripotent hemopoietic stem cells and its expression is maintained at high levels in maturing lymphocytes. Mice homozygous for a deletion of the Ikaros DNA‐binding domain lack committed lymphoid progenitors as well as mature T and B lymphocytes and natural killer cells (Georgopoulos et al., 1994). In addition to this apparent role in the early development of lymphoid progenitors, Ikaros is also required for later events during T cell maturation (Winandy et al., 1995). Mice heterozygous for this Ikaros mutation generate abnormal T cells. They develop lymphoproliferative disorders and ultimately die of T‐cell leukemias and lymphomas.
The Ikaros protein isoforms all share a common C‐terminal domain containing two zinc fingers to which different combinations of N‐terminal zinc fingers are appended. The N‐terminal zinc fingers are required for sequence‐specific DNA binding while the C‐terminal zinc fingers mediate homo‐ and heterodimerization among the Ikaros isoforms (Molnar and Georgopoulos, 1994; Sun et al., 1996). Homo‐ and heterodimerization of isoforms with a DNA‐binding domain greatly increases both their affinity for DNA and their transcriptional activity (Sun et al., 1996). Mutations that disrupt the C‐terminal zinc fingers prevent Ikaros proteins from engaging in higher‐order DNA interactions and from activating transcription (Sun et al., 1996). In addition, heterodimers which include one Ikaros isoform that lacks a DNA‐binding domain are transcriptionally inert. Hence, such isoforms can interfere with the activity of Ikaros isoforms that contain a DNA‐binding domain in a dominant‐negative fashion.
The last translated exon of the Ikaros gene shared by all of the Ikaros isoforms was targeted by deletion in the mouse germ line (Wang et al., 1996). This exon encodes the C‐terminal dimerization domain of the Ikaros proteins as well as an activation domain that mediates their effects in transcription (Sun et al., 1996). Mice homozygous for this mutation display a phenotype that is less severe than that caused by deletion of the Ikaros DNA‐binding domain. In the Ikaros C‐terminal‐mutant mice, fetal hemopoietic stem cells or their immediate progeny fail to enter the T and B lymphoid pathways. Throughout fetal life_and for a few days after birth_the thymus of these mice is devoid of a lymphoid compartment. In addition, pro‐B and pre‐B cells are not detected in the fetal liver of the Ikaros mutant embryos. However, during the first week after birth, increasing numbers of thymocyte precursors are seen in the thymus. These give rise to conventional αβ and some γδ T cells, but not to natural killer cells or any significant numbers of dendritic antigen‐presenting cells (APCs). B cells and their earliest described precursors are also absent from the spleen, the peritoneum and bone marrow. In addition, mice heterozygous for this Ikaros mutation are not obviously abnormal.
Given that the functionally inactive proteins generated by the Ikaros C‐terminal mutant locus are unstable and rapidly degraded in cells in which they are produced, this mutation is considered a null for Ikaros activity (Wang et al., 1996). Therefore, the complete lack of adult T lymphocytes in mice homozygous for the Ikaros DNA binding deletion can be explained by a dominant interfering effect of the encoded mutant Ikaros isoform on another protein. This factor must work in concert with Ikaros to specify at least T cell identity in the late fetal and postnatal hemopoietic system.
Since the zinc fingers in the Ikaros C‐terminal domain display strong homology to the C‐terminal zinc fingers of the Drosophila suppressor protein Hunchback (Tautz et al., 1987), it appears that this domain existed prior to the expansion of the vertebrate genome and may also be included in other proteins. These proteins would have the potential to interact with Ikaros proteins when co‐expressed and may act in concert during lymphocyte differentiation. Such interactors would be candidate targets for the dominant‐negative activity of the truncated Ikaros isoforms. To investigate this possibility, we used degenerate oligonucleotides to amplify the C‐terminal zinc finger domain from the mouse genome. Aiolos was identified as a homolog of Ikaros whose expression is restricted to the lymphoid lineage. The Aiolos protein shows extensive homology to the largest Ikaros isoform, Ik‐1, throughout the DNA binding and C‐terminal domains. Aiolos homomeric complexes are potent transcriptional activators while heteromers between Aiolos and different Ikaros isoforms range in activity from slightly less potent to transcriptionally inert. Unlike Ikaros, Aiolos is not expressed in fetal hemopoietic sites or in the early fetal thymus. Aiolos mRNA is detected in the late fetal thymus and in the adult lymphoid organs. Within adult hemopoietic progenitors, Aiolos is not expressed in pluripotent stem cells but is detected at low levels in multipotent progenitors. Its expression is dramatically up‐regulated as these progenitors become more restricted into the T and B lymphoid pathways.
We propose that Aiolos and Ikaros act in concert during lymphocyte development in the late fetal and postnatal hemopoietic system. In the absence of Ikaros, partial overlap in function between the two genes may allow for T cell, but not B cell, specification by Aiolos.
Functional domains are conserved between Aiolos and Ikaros
PCR with degenerate primers from the domain conserved between Ikaros and Hunchback identified several genes in the mouse genome which encode a pair of zinc fingers similar to those that mediate dimerization of the Ikaros proteins. Of these, the Aiolos gene exhibits the greatest similarity to Ikaros and its expression is restricted in the hemopoietic system. cDNAs derived from this gene contain an open reading frame encoding a 58 kDa protein similar in size and structure to the Ik‐1 isoform (Figure 1). Western analysis of thymic nuclear extracts with antisera raised against Aiolos identifies a single protein that migrates at a similar rate to that of the Ik‐1 isoform (Figure 1B). The two proteins share an overall 70% similarity. Four blocks of sequence are particularly well conserved. The first encodes the zinc finger modules contained in the Ik‐1 isoform which mediate DNA binding of the Ikaros protein (Molnar and Georgopoulos, 1994). The second block of conservation has not been characterized functionally. The third block of conservation is a domain required for transcriptional activation by Ikaros. The final block of conservation corresponds to the zinc fingers which mediate dimerization and were the basis of the screen. The 5′ region of the open reading frame is only weakly conserved.
Two highly conserved C‐terminal zinc finger motifs mediate interactions between Aiolos and Ikaros proteins
The C‐terminal zinc fingers which mediate dimerization of the Ikaros proteins and were the basis of the screen are well conserved between the two proteins (Figure 1). The ability of the Aiolos C‐terminal zinc finger domain to engage in protein interactions was tested in a yeast two‐hybrid assay (Fields and Song, 1989; Zervos et al., 1993). In this system, the Aiolos C‐terminal domain (Aio 500) interacted strongly with itself and with the equivalent Ikaros domain (Figure 2B; Aio 500 and Ik 500). It also interacted in a similar fashion with the full‐length Aiolos and Ikaros proteins expressed by the recombinant pJG prey vectors (Figure 2B, Aio and Ik‐1). This Aiolos zinc finger domain, however, did not interact with mutant Ikaros proteins which contain amino acid substitutions in their C‐terminal zinc fingers that disrupt their ability to dimerize (Figure 2B, Ik‐1 M1, M2 and M1 + M2; see also Sun et al., 1996). In a similar fashion, the equivalent Ikaros bait (Ik 500) interacted with recombinant prey proteins that contained either the C‐terminal domain of Aiolos or Ikaros or the full‐length proteins (Figure 2C, Aio 500, Aio 800, Aio, Ik 500 and Ik‐1). In this assay, the affinities of Aiolos for itself or Ikaros are similar and indistinguishable to that of Ikaros for itself. Hence, formation of heterodimers between these two proteins would be expected when they are co‐expressed.
In vivo physical interactions between the Aiolos and Ikaros proteins
Complexes between endogenous Aiolos and Ikaros proteins can be detected in lymphocytes by co‐immunoprecipitation. Immunoprecipitation of nuclear extracts from thymocytes with an affinity‐purified antibody directed against Aiolos co‐precipitates Ikaros proteins (Figure 3A, lane 2). Interactions between these proteins is mediated by their C‐terminal zinc fingers. When co‐transfected into fibroblast cells with epitope‐tagged (Brizzard et al., 1994) Aiolos, Ikaros protein is co‐precipitated with an antibody to the tagged Aiolos protein (Figure 3A, lane 3). Point mutations in the zinc finger domain which prevent Ikaros protein interactions (Sun et al., 1996) also prevent co‐precipitation of Aiolos and Ikaros proteins in this assay (Figure 3A, lane 4).
The interaction between these proteins can also be observed directly (Figure 3B). Unlike most Ikaros isoforms, the Ik‐6 isoform lacks a DNA‐binding domain and is normally found in the cytoplasm (Figure 3B, panel 3). When co‐expressed with Aiolos, Ik‐6 is detected in the nucleus (Figure 3B, panel 4). Immunofluorescence reveals a punctate pattern of nuclear staining for both Ikaros and Aiolos proteins (Figure 3B, panels 5–7). This pattern is similar to that of RING finger proteins like PML and members of the polycomb group (Messmer et al., 1992; Borden et al., 1995). When Aiolos is co‐expressed with an Ikaros isoform that is localized to the nucleus, both proteins are detected within the same nuclear speckles. Hence, Aiolos dimerizes with Ikaros proteins and localizes to the same discrete regions within the nucleus.
The nuclear speckle staining of endogenous Aiolos proteins is also detected in thymocytes and peripheral T and B cells (Figure 3B, panel 9 and data not shown). A very similar pattern is observed with Ikaros proteins in these primary lymphocytes (Figure 3B, panel 8).
Conserved function of the N‐terminal zinc finger DNA‐binding domain in Aiolos and Ikaros proteins
A second block of conservation spans the N‐terminal zinc finger domain of Ikaros which mediates its DNA binding (Molnar and Georgopoulos, 1994). Contacts between DNA and the α‐helical region in the C‐terminal half of Krüppel‐like zinc fingers are important in determining the sequence specificity of these interactions (Lee et al., 1989; Pavletich and Pabo, 1993). These regions are perfectly conserved between Aiolos and Ikaros (Figure 1) and both proteins are capable of binding the same DNA sequences with similar affinity. As demonstrated by EMSA, both Aiolos and Ikaros proteins form similar high‐affinity complexes with an oligonucleotide containing a binding site for the Ik‐1 protein (Figure 4A). Furthermore, competition with specific and mutated oligonucleotides show that these proteins have similar affinities for this binding site. Hence, Aiolos and Ikaros can, in principle, compete for target sites in the genome.
Alternative splicing leads to the production of multiple Ikaros isoforms with different combinations of N‐terminal zinc fingers appended to the common C‐terminal domain. The messages encoding these isoforms are readily detectable by PCR using primers from exons 2, 3 and 7 of the Ikaros gene (Figure 6‐Ik; see also Molnar and Georgopoulos, 1994). In contrast, PCR analysis performed with analogous primers derived from the Aiolos cDNA detected a single cDNA species (data not shown and Figure 6, Aio). Indeed, analysis with primers designed from Aiolos sequence corresponding to each of the exons in Ikaros failed to detect alternatively spliced products (data not shown). Western analysis of thymic nuclear extracts revealed a single protein which reacts with an antibody raised against Aiolos (Figure 1B).
Aiolos is a more potent transcriptional activator than Ikaros
Although the third block of conservation has not been characterized functionally, the fourth conserved block is a domain required by Ikaros for transcriptional activation (Sun et al., 1996). This activation domain is composed of a stretch of acidic amino acids followed by a stretch of hydrophobic residues, both of which are required for its full activation potential.
Despite the similarity between Aiolos and Ikaros in this region, Aiolos is a stronger activator in mammalian cells. When co‐transfected into fibroblast cells with a tkCAT reporter construct under the control of four copies of a single high‐affinity Ikaros binding site, Aiolos stimulated CAT activity by 25‐ to 50‐fold [Figure 4B, Aio (5) and Aio (10)]. In contrast, Ik‐1, the strongest transcriptional activator of the Ikaros family (Molnar and Georgopoulos, 1994), elicited a 12‐ to 25‐fold increase in expression in this assay [Figure 4B, Ik‐1 (5) and (10)]. Therefore, Aiolos homodimers can compete with Ikaros homodimers for binding sites and can stimulate transcription to higher levels. The net activity of resulting Ikaros–Aiolos mixtures was estimated by co‐expressing both proteins in fibroblasts at different ratios. These data suggest that Ikaros–Aiolos heterodimers have transcriptional activity which is intermediate between the activities of Aiolos and Ikaros homodimers [Figure 4B, Aio (10), Aio (5) + Ik‐1 (5), Ik‐1 (10)].
Ikaros isoforms which lack a DNA‐binding domain interfere with the transcriptional activity of Aiolos proteins when both are expressed in the same cell (Figure 4B, Aiolos + Ik‐6). Similar results were obtained when Ikaros isoforms with and without a DNA‐binding domain were co‐expressed (Sun et al., 1996). Heterodimers formed between these two functionally distinct Ikaros isoforms do not bind DNA and consequently cannot activate transcription (Sun et al., 1996). The dramatic decrease in Aiolos activity is most probably due to the formation of such functionally inactive Aiolos–Ikaros heterodimers. Transfection with equimolar amounts of Aiolos and an Ikaros isoform that lacks a DNA‐binding domain (Ik‐6) leads to the 65% reduction in CAT activity expected if Aiolos–Ik‐6 heterodimers are transcriptionally inert. Addition of higher levels of Ik‐6 further reduces transcription of the reporter gene. This effect is specific for the interfering isoform, since addition of similar amounts of an Ikaros isoform with an intact DNA‐binding domain leads to a linear increase in transcriptional activity [Figure 4B, Aio 5 + Ik‐1 (5–15)].
Aiolos expression is restricted to the lymphoid system
In the adult mouse, Aiolos transcripts are detected exclusively in lymphoid tissues and are ∼9 and 4.5 kb in size (Figure 5A). Aiolos expression levels are highest in the spleen, progressively lower in the thymus and bone marrow, and undetectable in non‐lymphoid tissues. The spleen is largely populated by mature B and T lymphocytes, while the majority of cells in the thymus are immature CD4+/CD8+ thymocytes which are in the process of rearranging their T‐antigen receptors. In the bone marrow, ∼25% of the cells are pre‐B cells at a stage of differentiation comparable with that of double‐positive thymocytes, while the remainder are predominately erythroid and myeloid precursors. Aiolos mRNAs are not detected in the bone marrow of Ikaros dominant‐negative mutant mice, the marrow being comprised largely of erythroid and myeloid cells and lacking detectable numbers of committed lymphoid precursors. These observations suggest that Aiolos is expressed predominantly in precursors of the B and T lineage, and is up‐regulated upon their terminal differentiation.
A detailed examination of the expression of Ikaros and Aiolos during embryonic development is consistent with this interpretation. In situ hybridization to embryo sections suggests that Ikaros is expressed at the earliest stages of hemopoiesis and prior to the development of committed lymphoid precursors. It is found in the day 8 yolk sac and in the hemopoietic fetal liver from E 9.5 of gestation (T.Ikeda, unpublished results; also Georgopoulos et al., 1992). In the thymus, it is present from the time that this organ is populated with the first wave of lymphoid progenitors (Havran and Allison, 1988; Georgopoulos et al., 1992). In contrast, Aiolos is not detected in the fetal liver, or in the early fetal thymus (Figure 5B, panel 4 and data not shown). This suggests that Aiolos is not expressed in hemopoietic progenitors that give rise to erythroid and myeloid precursors in the fetal liver, or in the lymphoid progenitors of epidermal γδ and conventional αβ T cells which seed the early fetal thymus and give rise to mid‐gestation fetal thymocytes (Havran and Allison, 1988, 1990; Raulet et al., 1991). RT–PCR analysis of RNA prepared from skin dendritic γδ T cells also failed to detect Aiolos, but readily detected Ikaros messages (data not shown). Aiolos is first detected in the mid‐ to late‐gestation thymus (E16), which is largely populated by double‐positive thymocytes and at this stage it is expressed at lower levels than Ikaros (Figure 5B, panels 5 and 6). However, it is strongly up‐regulated in the prenatal (E19) and postnatal thymus, at a time in development when the progeny of the later waves of T‐cell differentiation comprise the major population of this lymphoid organ (Figure 5B, panels 11 and 12). In contrast to Ikaros, Aiolos is not detected in the developing nervous system (data not shown).
To characterize further the relative expression of Aiolos and Ikaros during lymphocyte ontogeny, RNA from sorted lymphoid populations of wild‐type and mutant mice were analyzed by RT–PCR. Ikaros transcripts were readily detected in a population that is enriched for pluripotent and self‐renewing stem cells that give rise to the lymphoid and myeloid/erythroid lineages (Figure 6, lane 13, Sca‐1+/c‐kit+; also Spangrude et al., 1989; Van de Rijn et al., 1989; Okada et al., 1992). It is also expressed at high levels in the more committed myeloid and erythroid precursors (Figure 6, lane 12, Sca‐1−/c‐kit+; also Okada et al., 1992). In contrast, Aiolos expression was not detected in either of the stem cell or in the erythroid and myeloid‐enriched progenitor compartments (Figure 6, lanes 13 and 12). Low amounts of Aiolos were detected upon prolonged exposure of the RT–PCR reactions in the Sca‐2+ multipotent progenitors (Wu et al., 1991) that contain cells with a potential to differentiate along the lymphoid lineage (Figure 6, lanes 15 and 16, lin−/Sca‐1+/c‐kit+/Sca‐2dl/br). Similar exposures failed to detect Aiolos in the pluripotent stem cell and in the erythroid/myeloid‐enriched progenitor populations (Figure 6, lanes 13 and 12). Significantly, low levels of Aiolos were also detected in the bone marrow of the Ikaros dominant‐negative homozygous mutant mice (Figure 6, lane 11; also Georgopoulos et al., 1994). These mice lack all definitive lymphoid precursors as well as more mature lymphocytes. However, their bone marrow populations may contain multipotent progenitors with lymphoid characteristics which are blocked from any subsequent differentiation along the lymphoid lineage. No expression was detected in the spleen of these mice, even upon prolonged exposure (Figure 6, lane 8).
Committed T cell precursors progress from a double‐negative through a double‐positive stage to single‐positive thymocytes (Pearse et al., 1989; Godfrey and Zlotnik, 1993). Double‐negative precursors are rare in wild‐type mice. In Rag‐1‐deficient mice, which lack a component of the recombinase complex required for lymphocyte maturation, early B and T cell precursors arrested in development are present in the bone marrow and thymus respectively (Mombaerts et al., 1992; Shinkai et al., 1992). Low levels of Aiolos were detected in double‐negative pro‐thymocytes isolated from the Rag‐1 mutant thymus, whereas moderate levels of Ikaros were expressed in these cells (Figure 6, lane 4). Aiolos mRNA was strongly up‐regulated in double‐positive thymocytes and in the CD4 and CD8 single positive T cells derived from them (Figure 6, lanes 1–3).
In the B lineage, a similar pattern of Aiolos expression was observed. The pro‐B cells isolated from Rag‐1‐deficient mice expressed Ikaros, but very little Aiolos (Figure 6, lane 10). Pre‐B and B cells from wild‐type bone marrow expressed high levels of both the Ikaros and Aiolos genes (Figure 6, lane 9). Among cells sorted from the spleen, Aiolos was expressed at higher levels in B cells than in T cells, while Ikaros displayed the opposite pattern (Figure 6, lanes 5 and 6). Therefore, although Ikaros predominates during the early stages of T and B cell maturation, expression of Aiolos increases significantly during the intermediate stages of the T and B lineage and comes to exceed that of Ikaros in mature B cells.
It is believed that natural killer (NK) cells are of lymphoid origin and share a common precursor with T lymphocytes (Hacket et al., 1986; Rodewald et al., 1992). Expression of Ikaros and Aiolos was examined in the spleen of Rag‐1‐deficient mice which is enriched for NK cells (Mombaerts et al., 1992; Shinkai et al., 1992). Although Ikaros was abundantly expressed in Rag‐mutant splenocytes, significantly lower amounts of Aiolos were detected (Figure 6, lane 7). In Ikaros‐mutant mice, the spleen is populated by the non‐lymphoid branch of the hemopoietic system (Georgopoulos et al., 1994). Aiolos expression was not detected among these myeloid and erythroid cells (Figure 6, lane 8).
Thus, in contrast to Ikaros, which is present in significant amounts from the early pluripotent stem cell compartment, Aiolos is first detected at low amounts in progenitor populations with a lymphoid potential. Aiolos comes to predominate at the intermediate and late stages of T and B cell maturation.
Dominant‐negative (DN) and null (C) Ikaros mutations have shown independently that this gene is not only an essential regulator during the earliest steps of lymphoid lineage specification, but is also required for normal development of maturing thymocytes. However, the more severe phenotype of the DN mutant strain suggested that Ikaros heterodimerizes with another protein also required for lymphocyte differentiation. Mice homozygous for a deletion of the exons encoding the Ikaros zinc finger DNA‐binding domain lack detectable numbers of cells in any stage of the lymphoid lineage beyond the pluripotent hemopoietic progenitor. Animals heterozygous for this mutation exhibit defects in the T lineage, which lead to an abnormal accumulation of double‐positive thymocytes and ultimately result in T cell leukemias and lymphomas. In contrast, a null mutation generated by deletion of the last Ikaros exon results in a less severe phenotype. Heterozygotes are not obviously abnormal, and although homozygotes lack B cells, fetally derived αβ and γδ T cells, NK cells and dendritic APCs, they do have adult‐derived αβ and some γδ T cells. The dominant phenotype of the first mutation can be explained by the fact that the Ikaros isoforms generated by the mutant allele have an intact C‐terminal dimerization domain and can interact and interfere at the heterozygous level with the function of transcriptionally competent Ikaros proteins. However, in the hemopoietic progenitors of mice homozygous for this mutation, Ikaros mutant proteins must interfere with the activity of other transcription factors that share similar dimerization properties. Aiolos was identified as a gene encoding another transcription factor which is expressed in the lymphoid lineage and has the potential to interact with and modulate Ikaros activity. The lack of Aiolos expression during the early waves of fetal lymphopoeisis may explain the lack of cells derived from these waves in the Ikaros null mutant. Expression of Aiolos in early lymphoid progenitors of the late embryonic/postnatal waves may account for the partial rescue of the αβ T lineage in Ikaros null (C−/−) mice.
The Aiolos protein displays structural and functional similarities to the Ik‐1 isoform. The dimerization domains of these two proteins are extremely well conserved and there is no discernible difference between the affinity of these proteins for themselves or for each other. In a similar fashion, the DNA‐binding domains of these two proteins are nearly identical and either homodimer as well as heterodimers of the two can all bind the same DNA sequences with comparable affinity. Ikaros binding sites (TGGGAA) have been identified in many developmentally important lymphoid restricted genes, i.e. CD3δ, mb‐1, TDT, IL‐2R, etc. (Hahm et al., 1994; Molnar and Georgopoulos, 1994). Hence homo‐ and heteromeric complexes of these proteins may act to modulate the activity of the same set of target genes. The interplay between these proteins in the regulation of gene expression is further choreographed by additional Ikaros isoforms which can sequester either Ik‐1 or Aiolos in complexes which are transcriptionally inert.
Although Aiolos and Ikaros share strong similarities over their activation domain, Aiolos is a stronger activator in mammalian cells. The difference in activity of the two proteins can be accounted for by additional protein interactions that take place with domains of the Ikaros proteins which are not conserved in Aiolos (J.Koipally and K.Georgopoulos, unpublished results). Such protein interactions may specifically modulate the activity of Ikaros in mammalian cells during development, without affecting Aiolos directly.
The expression patterns of Ikaros and Aiolos suggest that variations in the relative levels of these proteins are important for the progression of a cell through the lymphoid lineage. Early in hemopoiesis, only Ikaros is expressed and Ikaros complexes are required and perhaps are sufficient to regulate the expression of genes that restrict a pluripotent stem cell to a progenitor with a lymphoid potential. As a consequence of these Ikaros‐mediated commitment events, Aiolos becomes expressed in primitive lymphoid progenitors and is incorporated in the Ikaros complexes. These Ikaros–Aiolos heteromeric forms are transcriptionally more active than the Ikaros complexes, and may regulate the expression of genes that control the transition to definitive T and B lymphocyte precursors. As Aiolos is up‐regulated in pre‐T (CD4+/CD8+) and pre‐B (B220/Igm−/+) cell precursors, the level of Aiolos in the Ikaros complexes increases and may allow for the later events in lymphocyte differentiation to take place. Finally, in mature B cells where Aiolos expression predominates, transcriptionally potent Aiolos homomeric complexes may control functions that are unique to these mature lymphocytes.
Therefore, normal progression through the T and B lineages may require the sequential expression of Ikaros–Ikaros, Ikaros–Aiolos and Aiolos–Aiolos complexes. Interference with Aiolos activity may affect normal lymphocyte maturation and function. In mice heterozygous for the DNA‐binding (dominant interfering) Ikaros mutation, defects in lymphocyte development are first observed in double‐positive thymocytes when Aiolos expression is normally up‐regulated. Since at this stage in differentiation Ikaros is expressed at higher levels than Aiolos, mutant Ikaros isoforms may readily sequester Aiolos proteins in inactive complexes which are unable to exert their function in T cell maturation. Although these dominant‐negative Ikaros isoforms are also expressed in B cells, defects in this mouse are limited to the T lineage. The different ratio of Aiolos to Ikaros mRNAs in B lymphocytes may result in insufficient mutant Ikaros proteins to titrate Aiolos and block its function in the B lineage.
Formation of transcriptionally potent Aiolos homomeric complexes in developing thymocytes may also have adverse effects on their maturation. Although mice homozygous for an Ikaros null mutation generate some αβ T cells, these cells differentiate abnormally (Wang et al., 1996). Lack of Ikaros expression and accumulation of Aiolos homomeric complexes may underlie some of the T cell defects, i.e. deregulated production of CD4 thymocytes, observed in the Ikaros null mice.
These studies on Aiolos expression and function, combined with our previous work on the Ikaros gene, suggest that both members of this gene family act in concert to regulate lymphocyte differentiation and function. Further studies on the in vivo and in vitro interactions between these proteins will help us elucidate their potential role in lymphocyte differentiation and immune responses.
Materials and methods
Cloning of Aiolos
Degenerate primers were designed from the Ikaros C‐terminal zinc finger dimerization region. They were used to PCR amplify a 65 bp band from genomic DNA. Amplified DNAs of the appropriate size were sequenced to identify Ikaros homologs and used as probes to screen mouse genomic libraries. Genomic clones obtained were analyzed for their sequence similarity to the last translated exon of the Ikaros gene. These genomic DNAs were also used to hybridize an adult mouse tissue RNA blot. A genomic clone which contained an open reading frame that displayed strong similarities to Ikaros exon 7 and hybridized to a 9 kb message in spleen and thymus was used as a probe to screen for its mouse homolog from a mouse spleen cDNA library. A number of partial cDNAs were obtained. Translation of a cDNA containing the entire coding region is displayed in Figure 1. Protein sequence comparisons were generated using the GCG Bestfit program (Genetics Computer Group, Inc.).
Separation of purified cell populations
B220+ and B220− populations were obtained from bone marrow and spleen of wild‐type C57BL/6 or RAG‐1−/− mice by magnetic cell sorting (Hardy et al., 1991). First, lymphocytes were enriched by centrifugation of total bone marrow or spleen cells through a layer of Lympholyte®‐M (Cedarlane Laboratories, Hornby, Canada). The enriched lymphocytes were washed twice with cold PBS/BSA (PBS supplemented with 1% BSA, 5 mM EDTA and 0.01% sodium azide), resuspended at a concentration of 107 cells/ml in PBS/BSA, and incubated at 6–12°C for 15 min with anti‐B220 MicroBeads (MACS). To monitor the purity of the positively selected cells and the flow‐through, fluorescein isothiocyanate (FITC)‐conjugated rat anti‐B220 antibody was added and incubated for a further 5 min. B220+ cells were separated using a MACS magnetic separation column (Miltenyi Biotec GmbH). FACS analysis of the resulting B220+ and B220− populations determined that these were 85–95% pure. Double‐positive and single‐positive thymic cell populations were obtained by flow cytometry of cells from thymuses of wild‐type C57BL/6 mice. Thymic cells were incubated for 30 min on ice with phycoerythrin (PE)‐conjugated anti‐CD4 and FITC‐conjugated anti‐CD8 antibodies (Pharmingen), after which they were washed and separated, using a Coulter sorter, into a single‐positive population, which included both CD4+CD8− and CD4−CD8+ cells, and a CD4+CD8+ double‐positive population (Wu et al., 1991). The single‐positive population was then further sorted into CD4+CD8− and CD4−CD8+ populations.
Hemopoietic stem and progenitor cell isolation
Bone marrow cell suspensions were prepared from 8‐ to 12‐week‐old C57Bl/6J mice by gentle crushing of whole femurs and tibias in a ceramic mortar using PBS containing 2% heat‐inactivated fetal bovine serum (PBS/2% FBS). Cells were layered over Nycodenz with a density of 1.077 g/ml (Nycomed, Oslo, Norway) and centrifuged for 30 min at 1000 g. The band of low‐density cells at the interface was removed, washed once in PBS/2% FBS, and resuspended in a cocktail of purified rat antibodies recognizing the lineage‐specific antigens CD11b/Mac‐1, CD45R/B220, Ly‐6G/Gr‐1, CD4, CD8 and Ter119 (Pharmingen, San Diego, CA). After a 30 min incubation on ice, the antibody‐coated cells were removed by two rounds of immunomagnetic bead depletion on a Vario MACS BS column (Miltenyi Biotec, Sunnyvale, CA) using a 23–G needle to restrict flow. The lineage‐negative cells were then stained with FITC‐conjugated D7 (anti‐Sca‐1) and PE‐conjugated anti‐c‐kit (Pharmingen) for 30 min on ice, followed by one wash in PBS/2% FBS containing 2 mg/ml propidium iodide (PI). Viable (PI‐negative) cells were sorted on a FACStar Plus (Becton‐Dickinson, San Jose, CA) directly into tubes containing 300 ml of Lysis Buffer RLT (Qiagen, Chatsworth, CA). Total RNA was prepared by homogenizing the samples (350 ml maximum) using QIA shredder columns and RNeasy spin columns (Qiagen). Samples of 5×104 cells were processed and the RNA was eluted in DEPC‐treated water in a final volume of 30 ml. Two‐color analysis of Sca‐1 and c‐kit revealed staining profiles identical to that reported by Okada et al. (1992). Based on these studies, Sca‐1+c‐kit+ (primitive repopulating stem cells) and Sca‐1−c‐kit+ (myeloid‐committed progenitors) were sorted. Lineage‐negative cells were also stained with anti‐Sca‐1‐FITC, anti‐c‐kit‐PE and anti Sca‐2‐Red 613 and sorted into Sca‐1+/Sca‐2−/l°, Sca‐1+/Sca‐2dull and Sca‐1+/Sca‐2bright.
RNA analysis and RT–PCR
The tissue distribution of the Aiolos gene was determined by Northern hybridization and by RT–PCR of total RNAs prepared from brain, heart, kidney, liver, thymus, spleen and bone marrow of an adult wild‐type mouse, and from the bone marrow and spleen of an Ikaros DN−/− mouse (Georgopoulos et al., 1994) and from the sorted populations described in the previous section. RNA purification and Northern analysis were performed as previously described (Georgopoulos et al., 1992). Up to 5 mg of RNA were reverse‐transcribed in a total volume of 25 ml, which included 1× first strand buffer (Gibco‐BRL), 4 mM DTT, 150 ng random hexamer primers, 0.4 mM of each deoxynucleotide triphosphate, 1 U Prime RNase inhibitor (5′ → 3′, Inc.) and 200 U Superscript II reverse transcriptase (Gibco‐BRL). RNA and primers, in a total volume of 12 ml, were heated to 65°C for 10 min before adding buffer, deoxynucleotides, DTT, RNase inhibitor and reverse transcriptase. The reactions were incubated at 37°C for 45 min, followed by an incubation at 42°C for 45 min. Finally, 1 U RNase H (Gibco‐BRL) was added, followed by an incubation at 37°C for 30 min. cDNAs were prepared from CD4+/CD8+ and CD4+, CD8+ sorted thymocytes, Rag‐1−/− thymocytes, B220+ cells from wild‐type bone marrow, B220+ cells from Rag‐1−/− bone marrow, B220+ and B220− cells isolated from wild‐type spleen, Ikaros−/− bone marrow and spleen, and Sca‐1−/c‐kit+ and Sca‐1+/c‐kit+ stem cell populations. cDNA from each reaction was used directly for radiolabeled PCR. Reactions included up to 4 ml of cDNA, 1× PCR reaction buffer (Boehringer‐Mannheim), 0.1 mg BSA, 100 ng each of 5′ and 3′ primers, 0.2 mM of each deoxynucleotide triphosphate, and 5 mCi each of [α‐32P]dATP and [α‐32P]dCTP (3000 Ci/mmol) in a total volume of 50 ml. Primers specific for Ikaros, Ex2F and Ex7R, have been previously described (Georgopoulos et al., 1994). Primers specific for Aiolos were:
AioA: ATCGAAGCAGTGCCGCTTCTCACC and
AioC: GTGTGCGGGTTATCCTGCATTAGC. Primers specific for GAPDH were:
GAPDHF: ATGGTGAAGGTCGGTGTGAACGGATTTGGC, and
Amplification parameters consisted of 95°C for 5 min, 60°C for 5 min, at which point Taq polymerase (Boehringer‐Mannheim) was added to each sample, followed by 27 cycles of 95°C for 15 s, 60°C for 20 s and 72°C for 30 s. PCR products were visualized by electrophoresis through an 8% polyacrylamide‐1× TBE gel, followed by autoradiography of the dried gels.
PCR analysis to determine splicing of Aiolos transcripts
Primer combinations AioC/AioA, Aio4F/AioA and Aio5F/AioA were used to examine the possibility of alternate splicing of the Aiolos mRNA. AioC anneals within exon 3, Aio4F within exon 4, Aio5F within exon 5 and AioA within exon 7. The primer sequences are the following:
AioC GTG TGC GGG TTA TCC TGC ATT AGC
AioF GTA ACC TCC TCC GTC ATA TTA AAC
Aio5F CGA GCT TTT CTT CAG AAC CCT GAC
AioA ATC GAA GCA GTG CCG CTT CTC ACC
In situ hybridizations
Sections were prepared from E12 to E16 embryos as previously described (Georgopoulos et al., 1992). These were incubated with Ikaros‐ or Aiolos‐specific [32P]UTP RNA sense and antisense probes at 51°C for 12–16 h. The Ikaros probe was 300 bp in size, generated from the 3′ untranslated region of its last exon. The Aiolos probe was generated from the first 330 bp of its last translated exon which show little homology to Ikaros sequences. Slides were washed with 0.5× SSC/0.1% SDS at 55 and at 65°C, dehydrated and dipped in diluted photographic emulsion (NBT2). Dipped slides were exposed for 4 weeks, developed, stained with hematoxylin and eosin, and analyzed by bright‐ and dark‐field illumination on an Olympus microscope. False color overlays of the exposed grains were generated using Adobe Photoshop.
Aiolos–Ikaros protein interactions
DNA fragments encoding full‐length Aiolos or its last 242 or 150 amino acids were cloned in frame to the LexA DNA‐binding domain in the bait pL‐202 and to the transcription activation domain B42 in the prey vector pJG4‐5 (Zervos et al., 1993). The recombinant Ikaros prey and bait vectors used have been previously described (Lei Sun et al., 1996). Combinations of Aiolos and Ikaros bait and prey vectors were transformed into the EGY48 yeast strain. EGY48 (MATa trp1 ura3 his3 LEU2::pLexAop6‐LEU2) has a Leu2 gene as well as the pJK103 plasmid harboring the lacZ gene under the control of two high‐affinity ColE1 LexA operators maintained under Ura3 selection. Growth of yeast cells on Ura− His− Trp− Leu−‐galactose plates and color development on Ura− His− Trp−‐X‐gal‐galactose plates were used to score Aiolos and Ikaros protein interactions.
Gel retardation assays
DNA binding assays were performed as previously described (Molnar et al., 1994). His‐tagged Aiolos and Ikaros proteins were produced in Escherichia coli BL21 using the pRSETvectors (Invitrogen) and purified by affinity chromatography on nickel–agarose columns. The purified proteins were tested for binding to the Ik‐BS1‐TCAGCTTTTGGGAATACCCTGTCA oligonucleotide which contains a high‐affinity Ikaros binding site (100 000 c.p.m./reaction = 1–5 ng). Competition assays were performed with Ik‐BS1 and with Ik‐BS8 TCAGCTTTTGGGggTACCCTGTCA oligonucleotides used at 5‐ to 100–fold molar excess.
Aiolos nuclear localization
Subcellular localization of Aiolos protein was determined upon its expression in NIH 3T3 fibroblasts (as described by Sun et al., 1996). The Aiolos protein was tagged with the Flag epitope and expressed from the CDM8 vector. A mouse monoclonal to the Flag epitope was used (Brizzard et al., 1994; Eastman Kodak). FITC‐conjugated goat anti‐mouse IgG (Boehringer‐Mannheim) was used at 1:2000 dilution for 2 h at room temperature. NIH 3T3 fibroblasts transfected with Aiolos and Ikaros expression vectors were stained with anti‐Flag and PE‐conjugated goat anti‐mouse and with anti‐Ikaros and goat anti‐rabbit IgG FITC sequentially. No cross‐reactivity between secondary antibodies was detected at the concentrations used. Cells were counter‐stained with Hoechst 33258 for 1 h in PBS at 1 mg/ml. The coverslips were mounted on slides with an anti‐quenching mounting solution (Vector Laboratories Inc.) and visualized on a Nikon fluorescence microscope with Nikon 60–100× oil objectives.
The reporter plasmids 4xIk‐BS1‐tkCAT or tkCAT alone were co‐transfected with recombinant CDM8 Aiolos or Ikaros vectors and with the pxGH5 plasmid at a ratio of 1:0.5:1 in NIH 3T3 cells (as described by Sun et al., 1996). Each transfection point was performed in triplicate or quadruplicate. 48 h after transfection CAT and GH assays were performed on cell lysates and supernatants. Part of the cell pellet was lysed in protein sample buffer and used for Western analysis to determine Aiolos and Ikaros protein expression in transfected fibroblast cells. The CAT activity base line was determined by co‐transfecting the CDM8 expression vector with the reporter plasmids. Less than 2% variability was detected between transfections performed in triplicate and <5% variability between experiments. The ability of Aiolos homo‐ and Aiolos–Ikaros heterodimers to stimulate CAT activity from the reporter plasmid 4xIk‐BS1‐tkCAT was subsequently determined. CDM8‐Aiolos was co‐transfected together with increasing amounts of either Ik‐1 or Ik‐6 expression vectors. In these transfection studies, CDM8 vector was used to supplement the amount of expression vector DNA to 20 mg. Transfection efficiencies were normalized by expression of growth hormone.
Immunoprecipitations were performed on cell lysates as described previously (Sun et al., 1996). Immunoprecipitations on untransfected lymphoid cells were performed with affinity‐purified antibody directed against Aiolos protein. Immunocomplexes were analyzed by Western blotting using a mouse monoclonal antibody specific for the Ikaros proteins. Precipitations of transfected complexes were performed with a mouse monoclonal to the Flag epitope. The recombinant expression vectors CDM8–Ik‐1 and Flag–Aiolos were used to transfect the epithelial 293T cell line.
The research was supported by a Cutaneous Biology Research Center (from Shiseido Co. Ltd) grant for pilot projects to B.Morgan and K.Georgopoulos and by an NIH RO1‐AI33062‐02 to K.G who is a Scholar of the Leukemia Society of America. We would like to thank A.Nichogiannopoulou for providing the Northern blot, and M.Bigby and S.Winandy for critical evaluation of the manuscript.
- Copyright © 1997 European Molecular Biology Organization